Bioacoustics Laboratory, Department of Cognitive Biology, University of Vienna, Vienna, Austria.
Behavioural Ecology Research Group, Department of Biology, Faculty of Science & Technology, Anglia Ruskin University, Cambridge, UK; Division of Biological Anthropology, University of Cambridge, Cambridge, UK.
J Voice. 2019 Jul;33(4):401-411. doi: 10.1016/j.jvoice.2018.01.003. Epub 2018 May 31.
Fundamental frequency (f) is often estimated based on electroglottographic (EGG) signals. Because of the nature of the method, the quality of EGG signals may be impaired by certain features like amplitude or baseline drifts, mains hum, or noise. The potential adverse effects of these factors on f estimation have to date not been investigated. Here, the performance of 13 algorithms for estimating f was tested, based on 147 synthesized EGG signals with varying degrees of signal quality deterioration. Algorithm performance was assessed through the standard deviation σ of the difference between known and estimated f data, expressed in octaves. With very few exceptions, simulated mains hum, and amplitude and baseline drifts did not influence f results, even though some algorithms consistently outperformed others. When increasing either cycle-to-cycle f variation or the degree of subharmonics, the SIGMA algorithm had the best performance (max. σ = 0.04). That algorithm was, however, more easily disturbed by typical EGG equipment noise, whereas the NDF and Praat's auto-correlation algorithms performed best in this category (σ = 0.01). These results suggest that the algorithm for f estimation of EGG signals needs to be selected specifically for each particular data set. Overall, estimated f data should be interpreted with care.
基频 (f) 通常基于声门电图 (EGG) 信号进行估计。由于该方法的性质，EGG 信号的质量可能会受到某些特征的影响，例如幅度或基线漂移、电源嗡嗡声或噪声。迄今为止，尚未研究这些因素对 f 估计的潜在不利影响。在这里，根据具有不同信号质量恶化程度的 147 个合成 EGG 信号，测试了 13 种估计 f 的算法的性能。通过已知和估计 f 数据之间差异的标准差 σ 来评估算法性能，以八度表示。除了极少数例外，模拟的电源嗡嗡声以及幅度和基线漂移都不会影响 f 的结果，尽管有些算法始终优于其他算法。当增加 f 的周期到周期变化或次谐波的程度时，SIGMA 算法的性能最佳（最大 σ=0.04）。然而，该算法更容易受到典型 EGG 设备噪声的干扰，而 NDF 和 Praat 的自相关算法在这一类中表现最好（σ=0.01）。这些结果表明，EGG 信号的 f 估计算法需要针对每个特定数据集进行专门选择。总体而言，应谨慎解释估计的 f 数据。
